Help: My map doesn't know how to add!
This will be a short post, because it’s all about the interactive map! Read it, work with it, and tell me what you think.
Outages
It’s July 3rd 2026, and NYC is undergoing a heat wave and resulting power outages. So lets go to the ConEd outage tracker map to find out where the outages are. All you need to know is that an outage happens in a location, and each outage can affect one or many customers — a single-family home vs a huge apartment building complex.
And what’s going on? We see markers. Great. What do they tell us?
Do they correspond to the location of outages? No, not directly
Do they tell me how bad the outage is? No: The number is outages, not customers
Can we at least know the number of customers? Only when zoomed in and the dots display. There’s a legend for that case, but the main usage of the map happens when zoomed out, so it’s not very useful.
And if there’s one thing I want to get across: We should never get used to something like this. It’s easy to display it better.
A map understands areas
Areas can do math.
How? Let me talk you through a quick mock. You can interact with it at:
https://chrisgoldammer.com/outage_map/
(Image attribution: Carto, OpenStreetMap)
At the outset, you see all areas affected
Areas are color-coded by the number of customers affected
As you zoom in, the area become more granular, and colors adjust. So the worst area is always red.
When you see the actual locations, the area of the circle corresponds to the customers affected.
I’m going to claim (with just the right amount of arrogance) that this is a pure improvement. We’re not making big assumptions, we’re just clarifying the implicit assumptions that the current map makes.
Can’t this be better? Why not go with some natural areas like boroughs, neighborhoods, blocks? Yes, this would likely be better, but that’s not the key point I’m trying to make here. You can usually make a map better by really understanding the location, but that can get tricky, e.g. it’s enormous work to now integrate a new city. All I want to show here is that you can make a map better without requiring lots of data complexity.
This is … everywhere
So where is this useful? Lots of places. Let’s try to buy a home:
Attribution: Streeteasy and BBL Club, Google, OpenStreetMap, Mapbox
The first image is from StreetEasy. To me personally, this map doesn’t feel right, because it hides results from me: As I zoom in, I see more results. “Why aren’t you telling me how much is for sale” is going through my head.
The second image is from BBL Club (it’s not about sales but that’s let’s just pretend for exposition). It’s the other extreme of how one can present this, similar to the original outage map. And here I ask: “What does this circle even correspond to”?
There’s something in between: Show dots, algorithmically selected, like StreetEasy, but then also tell me the total. The background tells me how popular an area is. And I naturally will be guided towards the hotter areas.
Conclusion
Am I just taking an imperfect map and telling you that it can be better? Partially: Yes! I dislike when people suffer from having a UI they can’t understand, and I think it’s good to have better things.
But the bigger message is that even well-designed maps can usually be much more informative. Maps are weird and they haven’t undergone the type of clean testing that we have for other UI elements. There’s lots of stuff we can potentially do with maps (aggregating, analysis, causal inferences) that we’re not used to, but it would be incredible to get this right.
I’m on it! And as I work on this, I’ll blog more about my attemps.